ChatGPT is now writing college essays, and higher ed has a big problem

AI in the classroom
(Image credit: Getty Images)

Higher education, we have a big, ChatGPT problem.

There are thousands of universities around the world, with their combined student populations numbering in the hundreds of millions – and at least a small percentage of them are now cheating with the help of ChatGPT

Even since I graduated with a Bachelor's Degree in, ahem, 1986, I've thought about returning to school to do a Master's or some other more prestigious degree. It was the prospect of the workload, though, that held me back; the countless hours of research and studying to produce even one paper. I know, it's weird to hear a writer say that, but there's a difference between reporting an impactful story and writing a college essay that, if you're lucky, your professor will read and, if you're even luckier, will give you better than a passing grade for.

That calculation, though, is now different thanks to the generative AI ChatGPT. Just this week, the BBC reported that a university student in the UK used the large language model (LLM) system to write one of his college essays and, after some editing, he received a better grade for it than he did for one he wrote himself.

The shame-free student even admitted that he plans on using ChatGPT for future college essays (unsurprisingly he didn't give his real name). Oddly, the university seems unconcerned, and appears convinced that the school body's skyrocketing visits to ChatGPT are mostly "from our research network."

Okay, you keep telling yourself that.

It's not that I think the majority of college students are cheaters. On the other hand, the stresses of juggling school, work, family, and other obligations can leave students feeling overwhelmed and looking for a leg up.

And students wouldn't be the only ones. People around the world are currently employing ChatGPT, Google Bard, and Bing AI to help them start and complete all kinds of projects. Sometimes it's just for inspiration; other times it's for huge swaths of usable text. So what if you can't 100% rely on them to get the facts straight or that, on occasion, they present falsehoods as fact?

It is not a cause for alarm; it is a call to action for universities...

Hofstra Univ. Provost Charlie Riordan

Most of us who've tried to use ChatGPT and other generative AIs for real work have found them sorely lacking. They can't code or create an OS, and they often lose their train of thought. I've already failed at using them for game programming and movie script writing.

On the other hand, ChatGPT gets better all the time – and what student wouldn't want to invest $20 a month in GPT-4-powered ChatGPT+, a drop in the bucket compared to the $25,000 or more a year they're already paying for a college education.

To sir, with AI love

I am not going back to college, but as an experiment I found a pair of Earth Sciences college courses at Princeton University, and asked ChatGPT to write essays that I could ostensibly hand in as coursework. I then emailed the results for each to the professors teaching those courses.

This was my first prompt:

"I need a 1,000 word essay for Princeton University's Environmental Fluid Mechanics about 'Determinants of Acute and Chronic Heat Stress in Urban Streams'"

ChatGPT never hesitates, and started fast typing (it has to be well over 90 words per minute) almost the entire essay. The system's character count limit cut it off a couple of paragraphs from the end. No biggie, I simply asked ChatGPT to deliver the rest. Overall, the final essay read well, but I realized that something was missing. I asked ChatGPT to add a bibliography, since there's no way a college professor would accept an essay that doesn't cite sources.

Once again, ChatGPT didn't hesitate, and quickly listed 10 sources.

ChatGPT also created a smart 500-word essay on 'Global Physical Climatology,' and this time I made sure to ask for the bibliography up front. I did ask ChatGPT to make it "entertaining." I'm not sure the resulting essay was any more entertaining than it would have been otherwise. It was, though, clear and compelling.

Lessons unlearned

Even though my college years are decades behind me, I still have a relationship with my alma mater, Hofstra University, serving as part of the School of Communications advisory board. I try, where possible, to help journalism students achieve their dreams.

Now, I wonder how many of them are doing so with the help of ChatGPT.

As well as the aforementioned Earth Sciences essays, I also gave this prompt to ChatGPT, for an essay I could share with the lecturers at Hofstra:

"I need a 500-word essay on the power and principles of social media story telling in investigative journalism for a Masters Degree in Journalism Social Media Storytelling class. Please me it original and interesting. Also include a bibliography."

Again, ChatGPT obliged, and I sent the resulting essay to the Dean of Journalism.

As I write this, none of the professors at Princeton or Hofstra have commented on my ChatGPT essays. Perhaps it's because they're all on spring break. It might also be that they read the essays, and were too shocked and horrified to respond.

Naturally, colleges and universities will want to get ahead of this problem, although they won't have much success if they take the same attitude as the aforementioned university in the UK, and assume that most students are only looking at ChatGPT for research purposes; it's like believing people who claim they're looking at adult content to learn more about human anatomy.

What educators need right now is a system akin to the plagiarism tools they use today to help them spot cribbed content. Those tools may not work here though; when I spot-checked my essays, I had trouble finding any direct quotes from previously written works (as I said, ChatGPT just keeps getting better). So, professors will need a new tool that can recognize the telltale signs of AI-generated essays.

ChatGPT happened so fast and is so easy to use; what university student wouldn't at least try it, even if only once?

What are those signs? I'm not certain. Maybe it'll be the veracity test, since these systems can still sometimes insist on presenting fantasy as fact, and in the most convincing way; that's not something a human student would normally do. They may write poorly, leave facts out, or even misinterpret them, but wholesale and convincing fabrication is not a typical part of real-student tradecraft.

To say I have a sinking feeling about this would be an understatement. ChatGPT happened so fast and is so easy to use; what university student wouldn't at least try it, even if only once?

Sadly, I think most students are like that student in the UK. They don't see ChatGPT as cheating. Instead, they see it as another tool, like a library, encyclopedia, or flash cards. The fact that it's more powerful, and can deliver some or all of your essay, is immaterial. ChatGPT's best work product comes out of a collaboration between AI and its interrogator. The first drafts of these AI-generated college essays aren't handed over to professors. Students take bits and pieces from them, or feed some AI-generated content back to ChatGPT and ask for alterations.

I guess, in a way, these college students are learning something about collaboration, editing, and, I hope, fact-checking. Are these the right lessons? I don't know – but universities and educators better figure it out, and fast.

While not commenting on my ChatGPT-generated investigative journalism essay, Hofstra University Provost and Senior Vice President for Academic Affairs Charlie Riordan did, in an email to me, encapsulate what may be the current academic zeitgeist surrounding this topic:

“Much is being written every day on this topic with the range of responses running the gamut. It is not a cause for alarm; it is a call to action for universities to better understand the technology and its impact on higher education, student learning, workforce, etc. We have launched a task force to provide a framework for campus-wide conversations.”

TOPICS
Lance Ulanoff
Editor At Large

A 38-year industry veteran and award-winning journalist, Lance has covered technology since PCs were the size of suitcases and “on line” meant “waiting.” He’s a former Lifewire Editor-in-Chief, Mashable Editor-in-Chief, and, before that, Editor in Chief of PCMag.com and Senior Vice President of Content for Ziff Davis, Inc. He also wrote a popular, weekly tech column for Medium called The Upgrade.

Lance Ulanoff makes frequent appearances on national, international, and local news programs including Live with Kelly and Mark, the Today Show, Good Morning America, CNBC, CNN, and the BBC. 

Read more
AI
ChatGPT is the homework helper for more than a quarter of teens – and the trend is accelerating
AI Learning for kids
AI doesn't belong in the classroom unless you want kids to learn all the wrong lessons
An AI-generated image of the colosseum with slides coming out of it.
AI slop is taking over the internet and I've had enough of it
EDMONTON, CANADA - FEBRUARY 10: A woman uses a cell phone displaying the Open AI logo, with the same logo visible on a computer screen in the background, on February 10, 2025, in Edmonton, Canada
I asked ChatGPT to work through some of the biggest philosophical debates of all time – here’s what happened
Sad writer
ChatGPT just wrote the most beautiful short story, and I wonder what I'm even doing here
ChatGPT
ChatGPT wants to write your next novel, and readers and writers alike should be very worried
Latest in Artificial Intelligence
AI Automated Agents
I've tested all the best AI agents including ChatGPT Deep Research and Gemini - these are the 5 top automated artificial intelligence tools you can try right now
Google Gemini Robotics
Gemini just got physical and you should prepare for a robot revolution
ChatGPT Parenting
I use ChatGPT to help with parenting - here's 5 prompts you can use AI to keep the kids entertained
Gemini on a smartphone.
I used Gemini AI to declutter my Gmail inbox and saved myself 5 hours a week – here’s how you can do the same
Sad writer
ChatGPT just wrote the most beautiful short story, and I wonder what I'm even doing here
ChatGPT
ChatGPT wants to write your next novel, and readers and writers alike should be very worried
Latest in News
UK Prime Minister Sir Kier Starmer
UK PM says AI should soon replace civil servants
Eight Samsung TVs mounted to the wall showing different basketball games
Samsung is offering you 8 new TVs in one bundle for March Madness, in case you want to watch all games at once like a Bond villain’s lair
The Steam Logo on a mobile phone in front of a wall of games.
Today’s Steam Spring Sale features my absolute favorite game of all time - here's when the sale starts and all the key info
Apple iPhone 16 Pro Max REVIEW
The latest iPhone 17 Pro Max leak may have given us another look at its upcoming redesign
Half-Life running on a smartwatch
This Redditor installed a game engine on their smartwatch, and now it runs Doom, Quake, and Half-Life
Samsung Galaxy Z Fold 6
The Samsung Galaxy Z Fold 7 could be in line for a Galaxy S25 Ultra-level camera upgrade